This project focuses on the study of membranes, proteins and carbohydrates by molecular dynamics computer simulation. Progress is reported under each Aim listed above Aim 1. Understand Model Membranes. This Aim involved applying a recently developed method for calculating bending constants of lipid bilayers of moderate size; i.e., approximately 400 lipids. Simulations were carried out on bilayers consisting of 648 DPPC, DOPC, and DOPE. Essentially quantitative agreement with bending constants from flicker experiments was obtained for DPPC and DOPC. This is significant because measurements from pipette aspiration and x-ray diffraction yield bending constants that are substantially smaller for these bilayers. Bending constants for DOPE were approximately twice that of the monolayer values determined from experiments in the inverse hexagonal phase. This is significant because the relationship that the bilayer bending constant is twice that of the monolayer is frequently assumed, but agreement between simulation and experiment has not be demonstrated until now (Levine, et al, JACS). Aim 2. Develop Simulation Methodology. A CHARMM compatible additive force field for sphingomyelin was developed. Calculated deuterium order parameters agree well with experiment, indicated that lipid surface areas are correct. Strong hydrogen binding between the NH and OH of the ceramide groups of neighboring lipids in the bilayer lead to slow dynamics and, remarkably, positive spontaneous curvature (Venable et al, Biophysical Journal). Aim 3. Simulate Complex Membranes Three papers related to this Aim were published, each covering a different topic. Simulations of two amphipathic peptides (Piscidin 1 and 3) were carried out in bilayers composed of different lipids. The simulation conditions were developed to match those of solid state NMR experiments carried out by collaborator Professor Myriam Cotten at Hamilton College who spent a sabbatical in the spring of 2012 in the Lab. The simulations provided a detailed characterization of the surface bound state, including secondary structure, depth/orientation in the membrane, and a previously unrecognized kink at the central glycine (Perrin et al, JACS). These simulations have set the stage for exploring the inserted, or pore, states of these peptides, which is more related to their antimicrobial action. Simulations of the amphipathic peptide ArfGAP1 in DOPE bilayers explicitly showed positive curvature induction. While this behavior has been predicted by continuum elastic models, the induction is twice that of the CEM. The discrepancy is explained in terms of the additional presence of specific interactions described only by the molecular model, thereby highlighting the utility of simulations for understanding experimental systems (Sodt and Pastor, Biophysical Journal). Lastly, 10 microsecond simulations of the liquid disordered and liquid ordered phases of DPPC/DOPC/cholesterol yielded near quantitative agreement with experimental deuterium order parameters, again demonstrating the accuracy of the C36 lipid force field developed in the group. With this agreement as validation, the structure of the liquid ordered phase was further characterized. As opposed the stoichiometric ratio of DPPC/cholesterol postulated in earlier studies, the simulations revealed a substructure of hexagonally ordered DPPC chains surrounded by cholesterol and an interstitial region enriched with DOPC. The balance of cholesterol-rich to local hexagonal order is proposed to control the partitioning of membrane components into the liquid ordered regions. These have been frequently associated with formation of so-called rafts, platforms in the plasma membranes of cells that facilitate interaction between components of signaling pathways (Sodt et al, JACS).